Understanding Projected Uncertainties in the Northern Winter Climate: Role of Interhemispheric Sea Surface Temperature Gradient and Arctic Sea Ice Cover

This study attempts to better understand the inter-model differences in the projected northern winter future climate based on the forced response in the CMIP5 RCP8.5 scenario. Inter-model empirical orthogonal function (EOF) analysis reveals that the dominant uncertainties in the future projections of sea level pressure (SLP) are in the eastern-to-southeastern flank of both the Aleutian low and the Icelandic low. These uncertainties in SLP are linearly related to a sea surface temperature (SST) pattern that primarily represents the interhemispheric SST gradient while the related sea ice cover (SIC) pattern represents the total Arctic sea ice extent. Corresponding monthly-varying SST and SIC perturbation patterns are computed and experiments with two atmospheric-only general circulation models (AGCMs) are performed to assess the contributions of these perturbation patterns in driving the projected uncertainties in SLP (i.e., the inter-model regression patterns) across CMIP5 coupled models.

The atmospheric response to the SST and SIC perturbations has a zonal wavenumber-2 pattern, with opposite SLP responses over the North Pacific and the North American continent, and a dipole-like SLP response over the North Atlantic resembling the North Atlantic Oscillation (NAO). A Rossby wavetrain driven by the SST perturbation, through accompanying convection anomalies over the tropical central-eastern Pacific, explains well the atmospheric response in the North Pacific. Downstream of the Pacific, the SST-driven Rossby wavetrain is modulated by the SIC perturbation, suggesting that the SST and SIC perturbations jointly explain the SLP response in North America and the NAO-like pressure response. These results highlight the primary role of the interhemispheric SST gradient and the secondary role of the Arctic SIC in the projected uncertainties in the Northern Hemisphere SLP. We also present a detailed comparison between the atmospheric response and the CMIP5 inter-model regression, and discuss the role of atmospheric circulation in the projected uncertainties in surface air temperature and precipitation.